Allocation of Time and Consumption-Equivalent Welfare: A Case of - - PowerPoint PPT Presentation

Allocation of Time and Consumption-Equivalent Welfare: A Case of South Korea

IARIW-BOK Special Conference Ki Young Park Soohyon Kim (presenter)

School of Economics Yonsei University

April 27, 2017

Park and Kim April 27, 2017 1 / 39

Roadmap

1 Allocation of Time in South Korea ◮ Motivation ◮ Data ◮ Individual-level time use ◮ Household-level time use ◮ Summary 2 Measuring consumption-equivalent welfare ◮ Background and conceptual difficulties ◮ Case I: log utility ◮ Case II: non-separable utility ◮ Decomposition 3 Discussion Park and Kim April 27, 2017 2 / 39

What we did

We document the allocation of time in market work, nonmarket work, child care, leisure using 2014 KLIPS survey, the first and most detailed time use survey in South Korea. We measure household-level consumption-equivalent welfare that consider consumption, (quality of) leisure, life expectancy, etc. We show that welfare measures that rely solely on income or consumption may be incomplete and misleading.

Park and Kim April 27, 2017 3 / 39

Main findings

Time use in market work, nonmarket work, child care, and leisure

◮ Men work longer hours, but their extra work is well-compensated by

more leisure and less hours in nonmarket work and child care.

◮ Leisure hour is a luxury good, consistent with observations in US and

• ther advanced countries.

Consumption-equivalent welfare

Park and Kim April 27, 2017 4 / 39

Motivation

Understanding on how economic agents allocate their time helps explain the various aspects of economic activities.

◮ Ghez and Becker (1975): substitutability of market and nonmarket work ◮ Greenwood et al. (2005): home production and women’s labor market

participation

◮ Aguiar et al. (2016): video games and a recent decline in hours of

young and less-educated men in US

South Koreans work the second-longest hours among OECD countries, but with relatively low labor productivity.

◮ Huge implication on quality of life and economic welfare ◮ Lack of detailed micro-level data on time use in South Korea Park and Kim April 27, 2017 5 / 39

Data

KLIPS (Korean Labor and Income Panel Study)

◮ Host organization: Korean Labor Institute ◮ Annual panel data of 5,000 households living in urban area, starting

from 1998

◮ household/personal/additional survey

We use 2014 additional survey on “Time Usage and Quality of Life,” the first and most detailed time use data in South Korea 2004 vs. 2014 additional survey

◮ 2004 survey has far fewer questions and thus provides far less detailed

information.

◮ It reports only total market hours and leisure. Park and Kim April 27, 2017 6 / 39

Sample and Classifications

Respondents aged 25 through 65 that are neither students nor retirees we try to follow the classifications in Aguiar and Hurst (2007) as closely as possible to make our results comparable. Time use classification Activities included (1) core market work main and side job (2) total market work (1) + commuting + job search (3) nonmarket work house-keeping activities (4) child care parenting (5) leisure measure 1 leisure activities (6) leisure measure 2 (5) + sleeping + personal care (7) leisure measure 3 (6) + childcare (8) leisure measure 4 (7) + others

Park and Kim April 27, 2017 7 / 39

Individual-level time use

According to OECD statistics, an average employed South Korean tends to work 2,124 hours in 2014, second to Mexico among OECD countries.

◮ In our dataset, the annual working hours amount to 2,320 hours. ◮ The share of employee who work longer than 52 hours per week is

23.6%.

Controlling for demographic, job-related variables, and others, men work longer by 3.67 hours. However, their extra work is well compensated by less nonmarket work (12.2 hours less), less childcare hours (0.36 hours less), and longer leisure (8-9 hours).

Park and Kim April 27, 2017 8 / 39

Regression, Individual-level

Park and Kim April 27, 2017 9 / 39

Individual-level, weekly

Park and Kim April 27, 2017 10 / 39

Trends in time use over the decade: 2004-2014

According to OECD statistics, South Korea has the fastest shortening working time in OECD. We have 3 comparable time use categories from 2004 and 2014 survey: total market work, leisure measure 1, leisure measure 2 Trend in total market work

◮ Over the decade, hours in total market work decline by 2.8 hours. ◮ And this decline is more noticeable for employed women by 8.8 hours.

Trend in leisure

◮ Interestingly, leisure measure 1 does not change much. ◮ Leisure measure 2 increases by 10 hours (mostly from an increase in

sleeping and personal care)

Park and Kim April 27, 2017 11 / 39

Trend, 2004-2014

Park and Kim April 27, 2017 12 / 39

Household-level time use

Double-income family works longer, spend less hours in nonmarket work, enjoys less leisure, and spends less hours in child care. For single-income family, non-working spouse spends more time in nonmarket work and child care. Having an infant (age 0-6) reduces leisure while having a kid (age 7-18) does not much.

Park and Kim April 27, 2017 13 / 39

Household-level, weekly

Park and Kim April 27, 2017 14 / 39

Women in double-income family

Gender inequality in favor of men?

Park and Kim April 27, 2017 15 / 39

Park and Kim April 27, 2017 16 / 39

Leisure hours as luxury good

The higher one’s income, the less hours in leisure. Table 8 with expenditure on leisure

Park and Kim April 27, 2017 17 / 39

2015 American Time Use Survey vs. 2014 KLIPS

Average work hours of employed person: 7.6 hours vs. 9.0 hours Men works longer than women: 42 minutes vs. 60 minutes Share of men doing food preparation and cleanup: 43% vs. 22% Average time per day women spend doing homework: 52 minutes vs. 2.85 hours For households with children under age 6,

◮ women spend: 1 hour vs. 4 hours ◮ men spend: 25 minutes vs. 1 hour Park and Kim April 27, 2017 18 / 39

Background

Jones and Klenow (2016): consumption-equivalent measure, relative welfare level compared to a target country, such as US Merits

◮ easy to compare welfare level among groups since it is a cardinal index

like

◮ consumption ratio, preserve multi-dimensional aspect as an welfare

index

Park and Kim April 27, 2017 19 / 39

Concept of consumption-equivalent welfare: an example

Per capital GDP and consumption in France are just 67% and 60% of the US values, but consumption-equivalent measure that considers leisure, mortality, and inequality is equal to 92% of that in the US

◮ question: “how much would you have been happy if you were born in

France, not in the US?”

◮ answer: “I would have enjoyed 92% of happiness as much as I do in the

US, because I could have benefitted from lower inequality, lower mortality, and more leisure despite lower consumption and income.”

Park and Kim April 27, 2017 20 / 39

Conceptual Difficulties

We need at least one reference group, but in household-level analysis within a country, it is not easy to find a reference group

◮ we take the top 20% group in terms of income as a reference group ◮ better than arbitrarily picking up a household as baseline

Subtle difficulty in interpreting variables such as inequality or mortality

◮ easy to interprete σ2

i as an inequality measure of country i

◮ conceptually vague if i refers to a household or an income quintile in

the same country

◮ better to define σ2

i as uncertainty of a household income within an

income group

Park and Kim April 27, 2017 21 / 39

How to Calculate

Simple case of log utility, calculating λq where q denotes an income group Cq and ℓq denote a household’s annual consumption and a measure

• f leisure, respectively

u(Cq, ℓq) = ¯ u + log Cq + ν(ℓq) = ¯ u + log Cq − θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ . Park and Kim April 27, 2017 22 / 39

How to Calculate

Simple case of log utility, calculating λq where q denote the quintile based on income Cq and ℓq denote a household’s annual consumption and a measure

• f leisure, respectively

u(Cq, ℓq) = ¯ u + log Cq + ν(ℓq) = ¯ u + log Cq − θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ . Park and Kim April 27, 2017 23 / 39

Consumption (Cq): measured real consumption less housing and education expenditure as well as non-consumption payment such as social insurance fee Leisure (ℓq): quantity or quality of leisure measure 1 4 and housework hours maybe comprised depending cases Household treated as if it consists of an individual representing overall household charateristics

◮ a: the representative individual’s age ◮ Sq(a): average survival rate for a household in each income quintile Park and Kim April 27, 2017 24 / 39

Lifetime Utility

Household’s lifetime expected utility Uq = E ∞

• a=1

βaSq(a)

• ¯

u + log Cq − θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ

• Park and Kim

April 27, 2017 25 / 39

Lifetime Utility

Independently and lognormally distributed household consumption in each income group

◮ arithmetic mean cq and a variance of log consumption of σ2

q

◮ E(log Cq) = log cq − 1

2σ2 q.

β = 1 and ℓ as being deterministic, lifetime expected utility is U(cq, ℓq) = LEq

• ¯

u + log cq − θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ − 1

2σ2

q

• ◮ where life expectancy LEq = ∞

a=1 S(a)q

Park and Kim April 27, 2017 26 / 39

Consumption Equivalent Measure λq

Consumption-equivalent welfare for quintile q, λq from following equation U(cq, ℓq) = U(λqc5, ℓ5)

Park and Kim April 27, 2017 27 / 39

Calibration

The Frisch elasticity of labor supply, ǫ = 1

◮ household consists of employed as well as non-employed members ◮ Moon and Song (2016): ǫ = 0.99 with intensive and extensive margin

The weight on the disutility from working, θ = w(1 − ℓ)−1/ǫ = 12.8

◮ w: aftertax real income, (1 − ℓ): labor supply c: real consumption at

the period,

¯ u is the intercept of utility function, conceptual utility from human dignity

◮ ¯

u = U(value of life in KR) − U( ¯ C, ¯ ℓ)

Park and Kim April 27, 2017 28 / 39

Case 1: Log Utility Function

income quintile ratio 1 2 3 4 5 income ratio (yq/y5) 0.224 0.349 0.453 0.591 1 consumption ratio (cq/c5) 0.505 0.591 0.687 0.783 1 equivalent measures of welfare λlog leisure measure 1 0.541 0.650 0.679 0.762 1 λlog leisure measure 2 0.515 0.613 0.675 0.766 1 λlog leisure measure 3 0.515 0.612 0.676 0.766 1 λlog leisure measure 4 0.519 0.615 0.680 0.761 1

Park and Kim April 27, 2017 29 / 39

Quality of Leisure in Log Utility Function

Becker (1965) quantity and quality of commodities consumed is important especially on household’s time allocation problem

◮ time and related expenditure determine the quality of consumption

ℓ: quality of leisure

◮ composite good with leisure hours and related expenditures as input

ℓm = lα

m x1−α m

◮ lm: leisure measure (m = 1, 2, 3, 4), xm: related expenditure Park and Kim April 27, 2017 30 / 39

Quality of Leisure in Log Utility Function

income quintile ratio 1 2 3 4 5 income ratio (yq/y5) 0.224 0.349 0.453 0.591 1 consumption ratio (cq/c5) 0.505 0.591 0.687 0.783 1 equivalent measures of welfare λlog quality of leisure 1 0.278 0.368 0.466 0.613 1 λlog quality of leisure 2 0.208 0.292 0.399 0.548 1 λlog quality of leisure 3 0.208 0.292 0.399 0.547 1 λlog quality of leisure 4 0.209 0.292 0.400 0.544 1

Park and Kim April 27, 2017 31 / 39

Case 2: Non-Separable Utility Function

With a non-separable utility function, leisure and consumption are substitute: Uq = C 1−γ

q

1 − γ

• 1 + (γ − 1) θǫ

1 + ǫ(1 − ℓq)

1+ǫ ǫ

γ

◮ γ = 1.5, consistent λNS within range 1 to 4 ◮ ǫ = 1, θ = 12.8

Derivative of Uq by Cq and ℓq. ∂2Uq ∂Cq∂ℓq = (1 − γ)

negative positive

• γθ(1 − ℓq)

1 ǫ C −γ

q

• 1 + (γ − 1) θǫ

1 + ǫ(1 − ℓq)

1+ǫ ǫ

γ−1

◮ λNS is consistently lower than λlog. Park and Kim April 27, 2017 32 / 39

Case 2: Non-Separable Utility Function

income quintile ratio 1 2 3 4 5 income ratio (yq/y5) 0.224 0.349 0.453 0.591 1 consumption ratio (cq/c5) 0.505 0.591 0.687 0.783 1 equivalent measures of welfare λNS leisure measure 1 0.308 0.475 0.574 0.683 1 λNS leisure measure 2 0.318 0.483 0.577 0.686 1 λNS leisure measure 3 0.319 0.482 0.578 0.685 1 λNS leisure measure 4 0.323 0.487 0.583 0.681 1

Park and Kim April 27, 2017 33 / 39

Quality of Leisure in Non-Separable Utility Function

income quintile ratio 1 2 3 4 5 income ratio (yq/y5) 0.224 0.349 0.453 0.591 1 consumption ratio (cq/c5) 0.505 0.591 0.687 0.783 1 equivalent measures of welfare λNS leisure measure 1 0.204 0.308 0.421 0.560 1 λNS leisure measure 2 0.205 0.309 0.422 0.561 1 λNS leisure measure 3 0.204 0.308 0.421 0.560 1 λNS leisure measure 4 0.204 0.308 0.421 0.557 1

Park and Kim April 27, 2017 34 / 39

Decomposing the Equivalent Measures

Due to additivity, possible to decompose λ into the forces that determine welfare log λq = log cq − log c5 (1) + θǫ 1 + ǫ(1 − ℓ5)

1+ǫ ǫ −

θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ

(2) + LEq − LE5 LE5

• ¯

u + log cq − θǫ 1 + ǫ(1 − ℓq)

1+ǫ ǫ − 1

2σ2

q

• (3)

+ 1 2σ2

1 − 1

2σ2

q

(4) Four components;

◮ (1) differences in means of log consumption ◮ (2) utility from leisure ◮ (3) life expectancy ◮ (4) uncertainty in log consumption. Park and Kim April 27, 2017 35 / 39

Decomposing the Equivalent Measures

income quintile components of log λlog 1 2 3 4 5 (1) mean of log consumption

• 0.683
• 0.523
• 0.375
• 0.245

(2) utility from leisure 0.129 0.084 0.013

• 0.001

(3) life expectancy

• 0.109
• 0.051
• 0.031
• 0.021

(4) uncertainty 0.021 0.022 0.020 0.017

Park and Kim April 27, 2017 36 / 39

Discussion

Blundell et al. (2016) Chiappori and Meghir (2014) Borra et al. (2016) isolate selection effects using longitudinal data from Australia, UK and US. We find that selection into marriage by individuals with a higher taste for home-produced goods can explain about half of the observed differences in housework documented in the cross-sectional data. Division of labor vs. gender inequality

Park and Kim April 27, 2017 37 / 39

Future research direction

Explicit treatment of household production Criteria for judging within-household gender inequality

Park and Kim April 27, 2017 38 / 39